Google Search Console and SEO glossary

An impression is counted when a link to your site appears on a Google results page a user loads, usually even if they never scroll down to it.

Search Console counts impressions in Google Search, Discover and News. Inside carousels, expandable results and infinite-scroll views such as mobile image search, the item typically has to be scrolled into view or expanded first. Scrolling or paging away and back during one query does not add a second impression. In the chart, data is aggregated by property, so two of your pages shown for the same query count as one impression; grouped by page, each URL gets its own impression. High impressions with few clicks often point to a title or snippet problem, see how to find low CTR keywords.

In GSC Magic: The gsc_query_analytics tool returns impressions grouped by query, page, country, device, search appearance, date or hour, and the dashboard charts them per day.

Related: Clicks, CTR (click-through rate), Average position, Page (dimension)

Sources: support.google.com, support.google.com

A click is counted when a user clicks a result in Google Search, Discover or News that takes them to a page outside Google.

Clicks that keep the user inside Google, such as a query refinement that starts a new search, are not counted. Clicking a result, coming back and clicking the same link again counts once; clicking a different link counts again. Click data is credited to the canonical URL Google selected, not necessarily the URL the user landed on. Search Console clicks and GA4 sessions measure different things and rarely match, see GSC clicks vs GA4 sessions.

In GSC Magic: The dashboard overview shows daily clicks and a Top queries/pages card with the change in clicks against the previous period.

Related: Impressions, CTR (click-through rate), Session (GA4), Canonical URL

Sources: support.google.com, support.google.com

CTR is clicks divided by impressions, shown as a percentage for your whole site or for a single query, page, country or device.

CTR depends heavily on position and on the type of result shown, so compare it between queries or pages at similar positions rather than against a single target. It also depends on aggregation: when several of your pages appear for one query, CTR grouped by property is generally higher than CTR grouped by page, because the property gets one impression and keeps the click. If impressions hold steady but clicks fall, Google suggests checking whether your title and snippet still describe the page well.

In GSC Magic: report_ctr_outliers lists pages whose CTR is under half of your site's median CTR at the same position.

Related: Clicks, Impressions, CTR outlier (low CTR page), Average position

Sources: support.google.com, support.google.com, developers.google.com

Average position is the average of the topmost position your site or page held in Google Search results, counted only for searches where it got an impression.

If one query shows your site at positions 2, 4 and 6 and another at 3, 5 and 9, Search Console counts 2 and 3 and reports an average of 2.5. Each result element occupies one position: all links in a knowledge panel or an AI Overview share that element's position, and ads don't occupy a position. A result that was never seen, for example on page 3, records no position, which is why some cells show a dash. Position is calculated only for Google Search, not Discover or News, and your own search can show a different rank because results depend on location and history. Google recommends watching changes over time rather than reading a single value.

In GSC Magic: report_quick_wins uses average position to find query and page pairs ranking 4 to 20 with many impressions.

Related: Impressions, Striking distance keywords (quick wins), AI Overviews and AI Mode, Core update

Sources: support.google.com, support.google.com

A query is a search term that led to your site appearing in Google results; the Queries tab of the Performance report groups clicks and impressions by it.

The Queries table is never complete. Some queries are hidden for privacy (anonymized queries), and because of internal limits Search Console stores and shows only the most important rows. Query filters are not case-sensitive. For the Web: multimodal search type, such as Google Lens searches, no text query is available, so the dimension is missing. Read more in why Search Console hides some queries.

In GSC Magic: gsc_query_analytics returns up to 25,000 query rows per request through the Search Analytics API.

Related: Anonymized queries, 1,000-row limit, Branded vs non-branded queries, Page (dimension)

Sources: support.google.com, developers.google.com

The Page dimension groups Performance data by the URL a search result linked to after redirects, usually credited to the page's canonical URL.

When a user clicks a duplicate URL, Search Console credits the click to the canonical URL Google chose, which you can check with the URL Inspection tool. Choosing the Pages tab switches table aggregation from property to page, so each URL shown for a query gets its own impression. The chart stays aggregated by property, which is the usual reason the table total and the chart total differ.

Related: Canonical URL, URL Inspection, Query, Impressions

Sources: support.google.com, support.google.com

Search appearance is a Performance report dimension that groups data by the result type or feature your page appeared in, such as review snippets, videos or merchant listings.

The filter only lists types for which your site recorded impressions. Grouping by search appearance aggregates the table by page. Clicks are assigned to a URL, not to a URL plus feature, so a filtered click count can include clicks on a plain link to the same page. Some older types, such as FAQ and How-to rich results, are listed by Google as deprecated.

In GSC Magic: gsc_query_analytics can group or filter performance data by search appearance.

Related: Structured data and rich results, Search type, Page (dimension)

Sources: support.google.com

Search type is the Performance report filter that separates traffic from Web (text-based or multimodal), Image, Video and News searches in Google Search.

Web: text-based covers queries typed into the standard search bar; Web: multimodal covers searches made with images, including Google Lens, Circle to Search and Chrome image search. Image, Video and News cover the matching tabs. Discover and Google News have their own performance reports. Google recommends analyzing search types separately when investigating a drop, and the API's row limit applies per search type.

Related: Search appearance, 1,000-row limit, Core update

Sources: support.google.com, developers.google.com

In the Performance report, Country groups data by where the search originated and Device groups it by desktop, mobile or tablet.

Both dimensions are aggregated by property. Google states that requests without the query or page dimension, such as countries, devices and search appearance, are displayed and exported in full, so they are not affected by the daily row limit. Comparing desktop and mobile can show a dash for position when a page had no impressions on one device.

In GSC Magic: gsc_query_analytics accepts country and device as dimensions and filters.

Related: 1,000-row limit, Average position, Query

Sources: support.google.com, developers.google.com

Data freshness is how soon Search Console shows performance data; the newest numbers can be preliminary, still being collected, and are drawn with a dotted line.

The 24-hour view shows the last available 24 hours with a delay of a few hours, in hourly points and in your browser's time zone. Other views use Pacific Time and show complete days by default. In the Search Analytics API, dataState "final" (the default) returns only finalized data, while "all" includes fresh data and "hourly_all" adds the hourly breakdown. Google's API guide notes that data is typically available after 2 to 3 days.

In GSC Magic: The hosted sync refreshes the last four days every six hours, because Google revises recent numbers.

Related: 16-month data retention, Clicks, Impressions

Sources: support.google.com, developers.google.com, developers.google.com, developers.google.com

Anonymized queries are searches Search Console leaves out of its tables to protect privacy, because they weren't issued by more than a few dozen users over two to three months.

They still count in the chart totals unless you apply a query filter, so the sum of the table rows is usually lower than the total. Filters drop them entirely: queries containing a word plus queries not containing it add up to less than the unfiltered total. See why Search Console hides some search queries.

In GSC Magic: GSC Magic reads the same Search Analytics API, so it cannot recover anonymized queries either.

Related: Query, 1,000-row limit, Branded vs non-branded queries

Sources: developers.google.com, support.google.com

The Search Console interface displays and exports at most 1,000 rows per table; the Search Analytics API returns up to 25,000 rows per request, with pagination.

Google states the API's upper limit is 50,000 rows per day per site per search type, which may not be reached in all cases. The API returns 1,000 rows by default; rowLimit raises that to 25,000 and startRow pages through the rest. Requests without query or page dimensions return all data. For the most complete query list, Google points to the bulk data export to BigQuery. Details in the 1,000-row limit explained and the Search Console API guide.

In GSC Magic: gsc_query_analytics requests up to 25,000 rows per query, so an AI agent works with far more than the interface's 1,000 rows.

Related: Query, Anonymized queries, 16-month data retention, Search type

Sources: developers.google.com, developers.google.com, support.google.com

Search Console's Performance report covers the last 16 months; older days drop out unless you copy the data elsewhere through the API or bulk data export.

Sixteen months is enough to compare a recent drop with the same weeks a year earlier, which is the check Google suggests for seasonality. For longer comparisons, Google suggests pulling data with the Search Analytics API or bulk data exports and storing it in your own systems. See 3 ways to keep Search Console data beyond 16 months and how GSC Magic stores history.

In GSC Magic: Pro and Agency backfill up to 16 months on upgrade and never prune stored history; Free keeps 14 days.

Related: Seasonality, 1,000-row limit, Content decay

Sources: developers.google.com

Crawling is how Google discovers and downloads pages; Googlebot is Google Search's crawler, with a smartphone and a desktop version.

Google finds new URLs mainly through links from pages it already knows and through sitemaps, then fetches them and renders JavaScript with a recent version of Chrome. Because Google primarily indexes the mobile version of most sites, most requests come from Googlebot Smartphone. For Google Search, Googlebot fetches the first 2 MB of a supported file type (64 MB for PDFs). Blocking crawling with robots.txt does not stop a URL from appearing in results; use noindex for that. The Crawl Stats report shows request counts, response codes and availability problems. Google doesn't accept payment to crawl a site more often.

In GSC Magic: GSC Magic does not crawl your site; it reads Google's URL Inspection data and Bing's crawl statistics and crawl issues.

Related: Indexing, Crawl budget, robots.txt, XML sitemap

Sources: developers.google.com, developers.google.com, support.google.com

Indexing is the stage where Google analyzes a crawled page and stores it in the Google index; only indexed pages can appear in Google Search.

Google Search works in three stages: crawling, indexing and serving results, and not every page makes it through each stage. Google doesn't guarantee it will index a page even if it follows Search Essentials. The Page indexing report explains why URLs are not indexed, for example "Crawled - currently not indexed" (crawled, may or may not be indexed later, no need to resubmit) or "Discovered - currently not indexed" (found, but the crawl was rescheduled to avoid overloading the site). See Search Console not indexed: causes and fixes.

In GSC Magic: report_index_health reads the URLs in your sitemaps, inspects them with the URL Inspection API and groups them by coverage state.

Related: Crawling and Googlebot, URL Inspection, noindex, Canonical URL

Sources: developers.google.com, support.google.com

Crawl budget is the set of URLs Google can and wants to crawl on a site, set by two factors: the crawl capacity limit and crawl demand.

The crawl capacity limit keeps Googlebot from overloading your server; crawl demand depends on the site's size, update frequency, page quality and relevance. Each hostname has its own crawl budget. Google says its crawl budget guide is mainly for very large sites (around 1 million or more pages changing weekly), sites with 10,000 or more pages changing daily, or sites with many URLs in "Discovered - currently not indexed". If new pages are crawled the day they are published, crawl budget is not your problem.

Related: Crawling and Googlebot, XML sitemap, Indexing

Sources: developers.google.com

A canonical URL is the URL Google picks as the most representative from a set of duplicate pages; it is the version usually shown in search results.

Duplicates come from HTTP and HTTPS versions, mobile and desktop URLs, sorting and filtering parameters, or regional variants. You can state a preference with rel="canonical", redirects and sitemaps, but Google treats it as a hint, not a rule, and may choose a different URL. Search Console credits clicks and impressions to the canonical. Compare the user-declared and Google-selected canonical in the URL Inspection tool.

In GSC Magic: gsc_inspect_url returns both the declared and the Google-selected canonical for a URL.

Related: Redirect (301 vs 302), Page (dimension), URL Inspection, hreflang

Sources: developers.google.com, developers.google.com

noindex is a rule, set with a robots meta tag or an X-Robots-Tag HTTP header, that tells Google to drop a page from its search results.

Google has to crawl the page to see the rule, so the page must not be blocked by robots.txt; otherwise it can still appear in results if other pages link to it. Google does not support noindex inside robots.txt. After you add the rule, a page stays in results until Google recrawls it, which can take months for less important pages. If robots.txt blocks the page, URL Inspection reports indexing as allowed, because Google can't see the noindex rule.

In GSC Magic: gsc_inspect_url and report_index_health show when Google found a noindex rule on a URL.

Related: robots.txt, Indexing, URL Inspection

Sources: developers.google.com, support.google.com

robots.txt is a file at the root of a site that tells crawlers which URLs they may access; it controls crawling, not whether a page is indexed.

Google describes it as a way to manage crawler traffic and avoid overloading your server. A page blocked by robots.txt can still be indexed without a description if other pages link to it, so use noindex or password protection to keep a page out of Google. Don't block CSS or JavaScript files that Google needs to understand the page.

In GSC Magic: The dashboard runs a quick robots.txt and sitemap audit for each site.

Related: noindex, Crawling and Googlebot, XML sitemap

Sources: developers.google.com

An XML sitemap is a file listing the pages, images, videos and other files on your site so search engines can find and crawl them more efficiently.

A sitemap can also carry the last modification date and alternate language versions. It helps discovery but doesn't guarantee crawling or indexing. Google says small, well-linked sites (about 500 pages or fewer) may not need one, while large, new or media-heavy sites benefit most. For many new or changed URLs, Google recommends submitting a sitemap rather than requesting indexing one URL at a time. See sitemap submitted but pages not indexed.

In GSC Magic: gsc_list_sitemaps shows status and errors, gsc_submit_sitemap submits one (Pro and Agency), and bing_submit_sitemap sends it to Bing.

Related: Crawling and Googlebot, Indexing, robots.txt, IndexNow

Sources: developers.google.com

URL Inspection is the Search Console tool that shows Google's indexed version of a URL, why it is or isn't indexed, and can live-test a page.

It reports indexing status, the Google-selected canonical, last crawl, structured data found and a screenshot of the rendered page. The live test checks whether a page might be indexable but cannot predict which URL Google will treat as canonical. The URL Inspection API returns only the indexed version, has no live test, and allows 2,000 inspections per site per day. See how to inspect 500 URLs with the API.

In GSC Magic: gsc_inspect_url inspects one URL and gsc_inspect_urls_batch inspects up to 500 in one call within Google's quota.

Related: Request indexing, Indexing, Canonical URL, Soft 404

Sources: support.google.com, developers.google.com, developers.google.com

Request indexing is a button in the URL Inspection tool that asks Google to crawl a single URL; it doesn't guarantee the page will be indexed.

You must be an owner or full user of the property. There is a quota for individual requests, and requesting the same URL several times won't make it crawl faster. Crawling can take from a few days to a few weeks. For many URLs, Google recommends submitting a sitemap instead. Google's separate Indexing API is officially limited to pages with JobPosting or BroadcastEvent markup. See IndexNow vs Google Indexing API.

In GSC Magic: The optional gsc_request_indexing tool uses Google's Indexing API, which Google supports only for JobPosting and BroadcastEvent pages.

Related: URL Inspection, XML sitemap, IndexNow, Indexing

Sources: developers.google.com, developers.google.com

A soft 404 is a page that returns a 200 success code but looks like an error or empty page to Google, which reports it and doesn't index it.

Google recommends returning a 404 status code for pages that truly don't exist, and adding more information to pages that should stay so they don't look like errors. To see what Google sees, run a live test in URL Inspection and open the screenshot of the rendered page.

Related: URL Inspection, Indexing, Redirect (301 vs 302)

Sources: support.google.com, developers.google.com

A redirect sends visitors and crawlers from one URL to another; a 301 is permanent and signals the target should be canonical, a 302 is temporary.

With permanent redirects (301, 308, instant meta refresh) Google usually shows the target URL in results. With temporary ones (302, 303, 307) Google follows the redirect but doesn't use it as a canonical signal, so the source URL may stay in results. Server-side redirects are the most reliable; Google's crawlers follow up to 10 hops by default. Search Console credits clicks to the final URL after redirects.

Related: Canonical URL, Soft 404, Page (dimension)

Sources: developers.google.com, developers.google.com

hreflang is an annotation that tells Google which pages are language or regional versions of the same content, so it can show users the right one.

You can declare it in HTML link tags, HTTP headers or the sitemap; the three methods are equivalent. Each version must list itself and all other versions, using fully qualified URLs. Google doesn't use hreflang to detect a page's language. Pages in the same language for different regions are duplicates, so Google recommends combining hreflang with canonicalization.

Related: Canonical URL, XML sitemap, Country and device

Sources: developers.google.com, developers.google.com

Structured data is a standardized format, mostly using schema.org vocabulary and often written as JSON-LD, that describes a page's content to Google and can make it eligible for rich results.

Rich results are search results with enhanced display, such as review stars, recipe details or product prices. A page needs all required properties to be eligible, and markup that breaks Google's structured data guidelines can make it ineligible. Google suggests testing markup with the Rich Results Test during development and watching the rich result reports in Search Console after deployment. Rich result types also appear in the Search appearance dimension.

In GSC Magic: gsc_inspect_url reports the rich results Google detected on a URL.

Related: Search appearance, URL Inspection, CTR (click-through rate)

Sources: developers.google.com, support.google.com

Core Web Vitals are three Google metrics of real-user experience: Largest Contentful Paint (loading), Interaction to Next Paint (responsiveness) and Cumulative Layout Shift (visual stability).

Good thresholds are LCP within 2.5 seconds, INP of 200 milliseconds or less and CLS of 0.1 or less, measured at the 75th percentile of page loads, separately for mobile and desktop. A page passes when all three meet the target. Google says good Core Web Vitals align with what its core ranking systems reward, as part of page experience. Search Console's Core Web Vitals report shows how your pages perform.

In GSC Magic: cwv_get_origin, cwv_get_url and cwv_history read real-user Core Web Vitals, and the dashboard shows a Core Web Vitals card per site.

Related: LCP (Largest Contentful Paint), INP (Interaction to Next Paint), CLS (Cumulative Layout Shift), CrUX (Chrome UX Report)

Sources: web.dev, developers.google.com

LCP measures loading: the time until the largest image, text block or video in the viewport is rendered. Good is 2.5 seconds or less.

LCP is measured from when the user first navigated to the page, so it includes server response time, redirects and connection setup. Use the 75th percentile of page loads, split by mobile and desktop, to judge whether most visitors get a good experience.

In GSC Magic: The cwv tools return LCP at the 75th percentile with the share of good, needs-improvement and poor loads.

Related: Core Web Vitals, INP (Interaction to Next Paint), CLS (Cumulative Layout Shift), PageSpeed Insights and Lighthouse

Sources: web.dev

INP measures responsiveness: how long the page takes to visibly respond to clicks, taps and key presses. Good is 200 milliseconds or less, poor above 500.

INP observes the latency of all interactions during a visit and reports roughly the slowest one, ignoring outliers. It replaced First Input Delay (FID), which measured only the input delay of the first interaction. Long JavaScript tasks on the main thread are a common cause of poor INP.

Related: Core Web Vitals, LCP (Largest Contentful Paint), CLS (Cumulative Layout Shift)

Sources: web.dev

CLS measures visual stability: the largest burst of unexpected layout shifts during a page's life. Good is 0.1 or less, poor is above 0.25.

A layout shift happens when a visible element moves between rendered frames without user input, for example when an image without set dimensions or a late ad pushes text down. Shifts less than one second apart are grouped into a session window, and the largest window is the CLS score.

Related: Core Web Vitals, LCP (Largest Contentful Paint), INP (Interaction to Next Paint)

Sources: web.dev

CrUX is Google's public dataset of how real Chrome users experience websites, and the field data source behind Core Web Vitals in Google's tools.

Origins and pages are included only if they are publicly discoverable and have enough visitors for a statistically significant sample, so many low-traffic pages have no data. Google Search uses CrUX to inform the page experience ranking factor. PageSpeed Insights shows CrUX as its field data and falls back to origin-level data when a page has too few samples.

In GSC Magic: The cwv tools query CrUX for an origin or URL; pages without enough Chrome traffic return no data.

Related: Core Web Vitals, PageSpeed Insights and Lighthouse, LCP (Largest Contentful Paint)

Sources: developer.chrome.com

PageSpeed Insights reports a page's real-user CrUX data and runs Lighthouse, an open-source tool that audits performance, accessibility, best practices and SEO in a simulated lab.

Field data shows what real visitors experienced over time; lab data from Lighthouse is a single controlled run that is useful for debugging but can miss real-world bottlenecks. PSI's Core Web Vitals assessment uses field data: a page or origin passes when the 75th percentile of LCP, INP and CLS is Good. Lighthouse also runs in Chrome DevTools, from the command line or as a Node module.

In GSC Magic: psi_run_audit runs Lighthouse through PageSpeed Insights, and the dashboard's Core Web Vitals card can start an audit.

Related: CrUX (Chrome UX Report), Core Web Vitals, LCP (Largest Contentful Paint)

Sources: developers.google.com, developer.chrome.com

Striking distance keywords are queries where a page already ranks just outside the top results, often positions 4 to 20, with enough impressions to gain clicks.

This is an SEO practice term, not a Search Console metric. The usual method: filter the Performance report by query and page, keep rows with an average position just below the top spots and meaningful impressions, then improve the existing page rather than writing a new one. Remember that average position is an average, so check the trend. See how to find striking distance keywords and 7 SEO quick wins.

In GSC Magic: report_quick_wins returns query and page pairs in positions 4 to 20 with many impressions, sorted by potential extra clicks.

Related: Average position, Impressions, CTR outlier (low CTR page)

Sources: support.google.com

Keyword cannibalization is when two or more pages on your site compete for the same query, splitting impressions and clicks instead of one page ranking well.

It is an SEO practice term rather than a Google one. In Search Console it shows up as one query with impressions spread across several URLs, sometimes swapping places over time. Not every overlap is a problem: different intents can deserve different pages, and branded queries often show many URLs. Common fixes are merging pages with a 301 redirect, differentiating their intent or adjusting internal links. See keyword cannibalization in Search Console.

In GSC Magic: report_cannibalization lists queries where two or more of your pages each get a meaningful share of impressions.

Related: Query, Page (dimension), Canonical URL, Redirect (301 vs 302)

Sources: support.google.com

Content decay is a gradual, sustained decline in a page's search clicks over several months, as opposed to a sudden drop from a technical issue or update.

It is an SEO practice term. Typical causes are outdated information, newer competing pages or shifting search interest. Comparing several consecutive periods separates a steady slide from one bad week; checking the same months a year earlier rules out seasonality. See how to detect content decay.

In GSC Magic: report_content_decay lists pages whose clicks fell in each of the last three windows (28 days each by default).

Related: Seasonality, 16-month data retention, Core update

Sources: developers.google.com

A CTR outlier is a page or query whose click-through rate is well below what other results at the same position on your site get.

Because CTR falls with position, a low CTR only means something when compared at similar positions. Outliers often have a title or snippet that doesn't match the query, or compete with a rich result or AI Overview on the same results page. Google suggests reviewing the title and snippet when impressions hold but clicks drop. See low CTR keywords in Search Console.

In GSC Magic: report_ctr_outliers flags pages whose CTR is under half of your site's median CTR at the same position.

Related: CTR (click-through rate), Average position, Striking distance keywords (quick wins)

Sources: developers.google.com, support.google.com

Branded queries include your brand name, its misspellings or your unique products; non-branded queries are all other searches that show your site.

Branded queries usually rank higher and get more clicks, while non-branded queries show how new people find you. Search Console has a branded queries filter in the Performance report, classified by an AI-assisted system rather than a keyword list. Google says it's available only for top-level properties with enough queries and impressions and may misclassify some queries.

In GSC Magic: report_brand_split splits queries by the brand terms you provide and returns clicks, impressions and click share for each group.

Related: Query, CTR (click-through rate), Anonymized queries

Sources: developers.google.com, support.google.com

Seasonality is a recurring rise or fall in search demand at the same time each year, such as holidays, that changes traffic without any ranking change.

Google's guide to traffic drops suggests setting the Performance report to the last 16 months to see whether a drop repeats every year, and checking top queries in Google Trends to see whether demand fell across the web. If impressions dropped while position held, demand is the likely cause. See seasonality vs real ranking losses.

In GSC Magic: Because Pro and Agency keep stored history beyond 16 months, agents can compare the same period across more than one year.

Related: 16-month data retention, Content decay, Core update

Sources: developers.google.com

A core update is a significant, broad change to Google's search ranking systems, made several times a year and announced on the Google Search ranking updates list.

Core updates don't target specific sites. Google recommends waiting until the rollout is finished, then comparing a full week afterwards with a week before it began. A small drop, such as position 2 to 4, needs no drastic action; a large, sustained drop calls for an honest assessment of the whole site's helpfulness. Google notes that some changes take effect in days, others in several months. See the organic traffic drop checklist.

In GSC Magic: report_traffic_drops compares a period with the one before it and lists the pages or queries that lost or gained the most clicks.

Related: Average position, Seasonality, Search type

Sources: developers.google.com, developers.google.com

Bing Webmaster Tools is Microsoft's counterpart to Search Console: search performance, crawl and index data, URL submission and keyword research for Bing.

Its API gives programmatic access to traffic and ranking stats, link and keyword details, crawl stats, and URL and sitemap submission. Search Performance shows 16 months of history. AI Performance, introduced as a public preview in February 2026, shows how often your pages are cited in Microsoft Copilot and Bing's AI-generated answers. See the Bing Webmaster Tools API guide and Bing Webmaster Tools in GSC Magic.

In GSC Magic: With your Bing API key, GSC Magic adds Bing tools for query and page stats, crawl data, keyword research and URL submission, plus report_google_vs_bing.

Related: IndexNow, GEO (generative engine optimization), XML sitemap

Sources: learn.microsoft.com, blogs.bing.com, blogs.bing.com

IndexNow is an open protocol that lets a site instantly notify participating search engines, including Bing, Naver, Seznam.cz, Yandex and Yep, that a URL was added, updated or deleted.

You host a key file on your site to prove ownership, then send one URL or a batch to an IndexNow endpoint; a submission is shared with all participating engines. An HTTP 200 response only means the URL was received, not that it will be indexed. Google is not listed among the participating engines. See how to set up IndexNow with an AI agent.

In GSC Magic: indexnow_generate_key creates a key and indexnow_submit sends up to 10,000 URLs per call (Pro and Agency, with your IndexNow key).

Related: Bing Webmaster Tools, Request indexing, XML sitemap

Sources: indexnow.org, indexnow.org, indexnow.org

In Google Analytics 4, a session is a period during which a user interacts with your site or app; on the web it ends after 30 inactive minutes by default.

A session starts when someone opens the app or views a page while no session is active, and GA4 logs a session_start event. There is no maximum length, and the timeout can be changed in the data stream settings. Sessions are not the same as Search Console clicks, see why the numbers don't match and GA4 in GSC Magic.

In GSC Magic: ga_run_report queries GA4 sessions read-only, by any dimension such as landing page or session source.

Related: Clicks, Key event (GA4), Referral traffic, AI-assistant traffic

Sources: support.google.com

A key event is a GA4 event you mark as important to your business, such as a purchase or sign-up; key events were previously called conversions in Analytics.

Any collected event can be marked as a key event. Google now uses "conversion" for actions measured for Google Ads campaigns and bidding, so conversion and key event counts can differ.

In GSC Magic: ga_list_key_events lists the key events configured on a GA4 property.

Related: Session (GA4), Referral traffic

Sources: support.google.com

In GA4, Referral is the default channel for visitors who arrive through non-ad links on other websites or apps, such as blogs and news sites.

GA4 assigns sessions to default channels using the source and medium. Organic search results, including Google's AI Overviews and AI Mode, go to Organic Search, and recognized AI assistants have their own AI Assistant channel. Direct covers visits from a saved link or a typed URL.

Related: AI-assistant traffic, Session (GA4), Key event (GA4)

Sources: support.google.com

AI-assistant traffic is visits from links in chat assistants such as ChatGPT, Gemini, Claude or Copilot, which GA4 groups in its AI Assistant channel when it recognizes the referrer.

Since May 2026 GA4 sets the medium to ai-assistant and the channel to AI Assistant when the referrer matches a recognized assistant; the channel excludes Google's AI Overviews and AI Mode, which count as Organic Search. The channel depends on the referrer, so assistant visits that arrive without one are not counted there. Traffic is not the same as citations. See track ChatGPT and Perplexity traffic in GA4.

In GSC Magic: The dashboard counts GA4 sessions from AI assistant sources such as ChatGPT, Perplexity, Claude and Gemini; it measures visits, not citations.

Related: Referral traffic, Session (GA4), GEO (generative engine optimization), AI Overviews and AI Mode

Sources: support.google.com, support.google.com

MCP (Model Context Protocol) is an open-source standard for connecting AI applications such as Claude or ChatGPT to external data sources, tools and workflows.

The MCP documentation compares it to a USB-C port for AI applications: one standard way to connect many systems. An AI application (the host) runs one MCP client per connection, and each client talks to an MCP server that exposes tools, resources and prompts. Servers run locally (stdio) or remotely over Streamable HTTP. See what happens when Claude calls Search Console through MCP.

In GSC Magic: GSC Magic is a hosted MCP server at https://mcp.gscmagic.com/mcp with 70 tools for Search Console, Bing, IndexNow, Core Web Vitals and GA4.

Related: MCP server and MCP client, Connector (Claude), GEO (generative engine optimization)

Sources: modelcontextprotocol.io, modelcontextprotocol.io

An MCP server is a program that exposes tools and data to AI applications over MCP; an MCP client is the component inside the AI application that connects to it.

The host application, such as Claude, an IDE or ChatGPT, creates one MCP client per server, and each client keeps one connection. Servers offer tools the model can call, resources it can read and prompts. A local server usually serves one client over stdio; a remote server using Streamable HTTP can serve many clients. See what a Search Console MCP server is and the full tool list.

In GSC Magic: GSC Magic's remote server signs clients in with OAuth or accepts a gsm_ access token, which can be read-only and expire.

Related: MCP (Model Context Protocol), Connector (Claude)

Sources: modelcontextprotocol.io, modelcontextprotocol.io, modelcontextprotocol.io

In Claude, a connector links the assistant to an external tool or data source; custom connectors connect Claude to remote MCP servers over the internet.

Claude connects to a custom connector's server from Anthropic's cloud, not from your device, so the server must be reachable on the public internet. Anthropic notes that custom connectors can reach services it hasn't verified, so add only servers you trust. See connect Search Console to Claude and all supported clients.

In GSC Magic: GSC Magic is listed in Claude's connector directory and can also be added as a custom connector or, in ChatGPT, as a custom MCP server.

Related: MCP (Model Context Protocol), MCP server and MCP client

Sources: support.claude.com

GEO (generative engine optimization) is the practice of improving how often and how well content appears in answers generated by AI search engines and assistants.

The term was introduced in a 2023 research paper that defined generative engines as systems that gather and summarize sources with large language models. Microsoft describes Bing's AI Performance report as an early step toward GEO tooling. Google says the same SEO fundamentals apply to its generative AI features and there are no extra requirements to appear in them. See how ChatGPT, Claude and Perplexity choose sources and GSC Magic for GEO specialists.

Related: AI Overviews and AI Mode, AI-assistant traffic, Bing Webmaster Tools

Sources: arxiv.org, blogs.bing.com, developers.google.com

AI Overviews and AI Mode are Google Search features that answer with AI-generated responses and links to supporting web pages; their traffic counts in Search Console.

AI Overviews appear only when Google's systems judge them helpful; AI Mode handles longer, exploratory questions. Both may use query fan-out, running related searches to find supporting pages. To be eligible, a page must be indexed and eligible for a snippet; there are no extra requirements. In the Performance report an AI Overview occupies a single position, and a click on an external link counts as a click. A Generative AI performance report, rolled out worldwide by August 31, 2026, shows impressions in these features. See how to track AI Overviews and AI Mode.

Related: Average position, GEO (generative engine optimization), AI-assistant traffic, Impressions

Sources: developers.google.com, support.google.com, support.google.com, developers.google.com